1

SPICE gives a default value for the (collector) substrate leakage saturation current ISS. The default is ISS=0 but I would like to know the real value for an integrated vertical npn transistor. The transistor is an older high-voltage device used in 1970 until today with many analog chips. The 44 Volt npn in the following figure is what I am interested in using. This figure comes from the "Analysis and Design of Analogue Integrated Circuits", 5th edition, by Gray, Hurst, Lewis and Meyer. BTW, if anyone knows additional SPICE parameters for this device, please provide them or references to them. This book seems to be the only place to get this information which seems a bit strange.

44 Volt Vertical NPN

ISS is also known as Saturation Current for Injection and can have a default value = IS Amp. Here is where it fits in the Gummel-Poon model for both vertical and lateral devices (i.e, collector or base substrate leakage saturation current respectively).

enter image description here

Mark
  • 21
  • 5
  • I'm not an IC designer, but there may not be a "typical" value for integrated circuits, since they are built according to that particular chip's requirements. Even what you see there is for a narrow margin of them all. If modelling is all you need, then there's no need to over do it with accuracy, cince the models for transistors are approximations, anyway. Many times, using some (reasonable) value then hammering if needed can bring good enogh results. SPICE is not for verifying a circuit, it's for insight into its functionality; the breadboard is for verification. – a concerned citizen Feb 08 '21 at 18:30
  • Thanks, yes of course but that is not exactly what I am trying to do. It is not about using this npn in a circuit to approximately simulate the circuit. I will use many versions of this npn's throughout the circuit. It's about a method of getting these versions with a "basis" npn for each bjt in the IC. The "basis" is specified as an exact reference, the type of method is what's being tested against a target that is also specified to be exact. So this is more about testing a methodology with open source tools. – Mark Feb 08 '21 at 18:45
  • Well, if it matters, and if you have LTspice, you could try opening up `LM741.asc` or `LM78xx.asc` from the `My Documents/LTspiceXVII/examples/Educational` folder, that comes with the default installation. I've also seen other attempts at IC modelling using `2N3904/3906`, or similar, with surprising results. – a concerned citizen Feb 08 '21 at 18:50
  • They have to come from IC models for the rated supply and match the way these IC's were made. Some other reasons these cannot be used are that discrete transistors are very large, with huge parasitic capacitance's, and the simulated IC may become unstable and oscillate (e. g., CJC = 7 pF, CJE = 22 pF in the 2N2222A, two orders of magnitude above the data in my table). Second, these discrete transistors are with only three terminals. In ICs, it is important to take into account the fourth terminal (bulk), available in most versions derived from SPICE 3f5. – Mark Feb 08 '21 at 19:00
  • I know, just pointing out a few suggestions. At any rate, I don't know, but someone else might. – a concerned citizen Feb 08 '21 at 20:16
  • I don't think it's that strange that you're having trouble finding an ISS value. ISS was not a parameter in the original SPICE2 or SPICE3, and only the capacitance of the substrate junction was modeled. I believe it was first defined by HSPICE. Then other programs like PSpice, LTspice, and ngspice implemented the parameter in order to be cross-compatible. – Ste Kulov Feb 09 '21 at 21:03
  • It appears so Ste. I went with ISS=IS – Mark Feb 09 '21 at 23:28

0 Answers0